ON GENERALIZED MULTIPLE-INSTANCE LEARNING
International Journal of Computational Intelligence and ApplicationsPublished 1 March 2005
Stephen Scott, Jun Zhang, Joshua W. Brown
Citations68
SJR quartileQ3
SJR score0.28
SNIP0.47
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
A generalisation of the multiple-instance learning model in which a bag's label is not based on a single instance's proximity to a single target point, but on a collection of instances, each near one of a set of target points.
Abstract
We describe a generalisation of the multiple-instance learning model in which a bag's label is not based on a single instance's proximity to a single target point. Rather, a bag is positive if and only if it contains a collection of instances, each near one of a set of target points. We then adapt a learning-theoretic algorithm for learning in this model and present empirical results on data from robot vision, content-based image retrieval, and protein sequence identification.
Keywords
Computer ScienceBiochemistry, Genetics and Molecular Biology
NeurocomputingAdvances in neural information processing systems 7
22,300 Citations1997Krzysztof J. Cios, Mark E. Shields
Communications on Pure and Applied MathematicsOrthonormal bases of compactly supported wavelets
8,187 Citations1988Ingrid Daubechies
This work construct orthonormal bases of compactly supported wavelets, with arbitrarily high regularity, by reviewing the concept of multiresolution analysis as well as several algorithms in vision decomposition and reconstruction.
IEEE Transactions on Pattern Analysis and Machine IntelligenceComparing images using the Hausdorff distance
4,474 Citations1993D.P. Huttenlocher, G.A. Klanderman +1 more
Efficient algorithms for computing the Hausdorff distance between all possible relative positions of a binary image and a model are presented and it is shown that the method extends naturally to the problem of comparing a portion of a model against an image.
Biological sequence analysis probabilistic models of proteins and nucleic acids
3,654 Citations2010Richard Durbin
Artificial IntelligenceSolving the multiple instance problem with axis-parallel rectangles
2,886 Citations1997Thomas G. Dietterich, Richard H. Lathrop +1 more
Three kinds of algorithms that learn axis-parallel rectangles to solve the multiple instance problem are described and compared, giving 89% correct predictions on a musk odor prediction task.
IEEE Transactions on Pattern Analysis and Machine IntelligenceSIMPLIcity: semantics-sensitive integrated matching for picture libraries
2,181 Citations2001James Z. Wang, Jia Li +1 more
SIMPLIcity (semantics-sensitive integrated matching for picture libraries), an image retrieval system, which uses semantics classification methods, a wavelet-based approach for feature extraction, and integrated region matching based upon image segmentation to improve retrieval.
Support Vector Machines for Multiple-Instance Learning
1,384 Citations2002Stuart Andrews, Ioannis Tsochantaridis +1 more
The proposed extensions of the Support Vector Machine learning approach lead to mixed integer quadratic programs that can be solved heuristic ally and a generalization of SVMs makes a state-of-the-art classification technique, including non-linear classification via kernels, available to an area that up to now has been largely dominated by special purpose methods.
Machine LearningLearning Quickly When Irrelevant Attributes Abound: A New Linear-Threshold Algorithm
1,376 Citations1988Nick Littlestone
This work presents one such algorithm that learns disjunctive Boolean functions, along with variants for learning other classes of Boolean functions.
A Framework for Multiple-Instance Learning
1,183 Citations1997Oded Maron, Tomás Lozano‐Pérez
A new general framework, called Diverse Density, is described, which is applied to learn a simple description of a person from a series of images containing that person, to a stock selection problem, and to the drug activity prediction problem.
Information and ComputationExponentiated Gradient versus Gradient Descent for Linear Predictors
880 Citations1997Jyrki Kivinen, Manfred K. Warmuth
The bounds suggest that the losses of the algorithms are in general incomparable, but EG(+/-) has a much smaller loss if only a few components of the input are relevant for the predictions, which is quite tight already on simple artificial data.
StructureThioredoxin —a fold for all reasons
816 Citations1995Jennifer L. Martin
The thioredoxin fold is a characteristic protein structural motif that has been found in five distinct classes of proteins that have the common property of interacting with cysteine-containing substrates.
Journal of Machine Learning ResearchImage Categorization by Learning and Reasoning with Regions
633 Citations2004Yixin Chen, James Z. Wang
This paper presents a new learning technique, which extends Multiple-Instance Learning (MIL), and its application to the problem of region-based image categorization, and provides experimental results on an image categorizing problem and a drug activity prediction problem.
EM-DD: An Improved Multiple-Instance Learning Technique
617 Citations2001Qi Zhang, Sally A. Goldman
A new multiple-instance (MI) learning technique that combines EM with the diverse density (DD) algorithm, which provides a new framework for MI learning, in which the MI problem is converted to a single-instance setting by using EM to estimate the instance responsible for the label of the bag.
CogPrints (University of Southampton)Solving Multiple-Instance Problem: A Lazy Learning Approach
577 Citations2000Jun Wang
Multiple-Instance Learning for Natural Scene Classification
569 Citations1998Oded Maron, Aparna Lakshmi Ratan
It is shown that very simple templates are suu-cient, and that performance improves with more user interaction, and the Diverse Density algorithm which is a method of learning from ambiguous examples is discussed.
Artificial IntelligenceQualitative navigation for mobile robots
331 Citations1990Tod S. Levitt, Daryl T. Lawton
This work has developed a multi-level theory of spatial representation of the environment based upon the observation and re-acquisition of distinctive visual events, i.e., landmarks, that provides the theoretical foundations for a visual memory database that smoothly integrates available metric knowledge of relative or absolute angles and distances.
IEEE Transactions on Image ProcessingObject matching algorithms using robust Hausdorff distance measures
264 Citations1999Donggyu Sim, Oh‐Kyu Kwon +1 more
This work analyzes the conventional Hausdorff distance measures and proposes two robust HD measures based on m-estimation and least trimmed square which are more efficient than the conventional HD measures.
Content-Based Image Retrieval Using Multiple-Instance Learning
245 Citations2002Qi Zhang, Sally A. Goldman +2 more
The multiple-instance (MI) learning model is applied to use a small number of training images to learn what images from the database are of interest to the user.
A multi-resolution technique for comparing images using the Hausdorff distance
205 Citations2002D.P. Huttenlocher, W.J. Rucklidge
An efficient method of computing the Hausdorff distance based on a multi-resolution tessellation of the space is possible transformations of the model set, with emphasis on the case in which the model is allowed to translate and scale with respect to the image.
Lecture notes in computer scienceA Two-Level Learning Method for Generalized Multi-instance Problems
162 Citations2003Nils B. Weidmann, Eibe Frank +1 more
This paper investigates a generalized view of the MI problem where an "interaction" between instances in a bag determines the class label, and transforms an MI bag into a single meta-instance that can be learned by a standard propositional method.
Image database retrieval with multiple-instance learning techniques
160 Citations2002Yang Chen, Tomás Lozano‐Pérez
An approach for retrieving images from an image database based on content similarity using a multiple-instance learning method known as the diverse density algorithm, which tries to retrieve images with similar feature vectors from the remainder of the database.
International Conference on Machine LearningMulti instance neural networks
157 Citations2000Jan Ramon, Luc De Raedt
Multiple Instance Regression
121 Citations2001Soumya Ray, David Page
The goals of this work are to understand the computational complexity of the multiple instance regression task and develop an efficient algorithm that is superior to ordinary multiple regression when applied to multiple instance data sets.
Lecture notes in computer scienceAttribute-value learning versus inductive logic programming: The missing links
118 Citations1998Luc De Raedt
It is much more tractable to stick to the relational representation than to apply the sketched transformation, which provides a sound theoretical justification for inductive logic programming.
DSpace@MIT (Massachusetts Institute of Technology)Learning from ambiguity
116 Citations1998Oded Maron, Tomás Lozano‐Pérez
This thesis will examine one framework of learning from ambiguous examples known as Multiple-Instance learning, which has developed a measure called Diverse Density and algorithms for learning from multiple-instance examples and applied these techniques to problems in drug design, stock prediction, and image database retrieval.
Machine LearningA Note on Learning from Multiple-Instance Examples
115 Citations1998Avrim Blum, Adam Tauman Kalai
All concept classes learnable with one-sided noise, which includes all concepts learnable in the usual 2-sided random noise model plus others such as the parity function, are learnable from multiple-instance examples.
PAC learning axis-aligned rectangles with respect to product distributions from multiple-instance examples
89 Citations1996Philip M. Long, Lei Tan
BioinformaticsIdentification of novel multi-transmembrane proteins from genomic databases using quasi-periodic structural properties
84 Citations2000Junhyong Kim, Etsuko N. Moriyama +3 more
The QFC algorithm uses concise statistical variables as the 'feature space' to characterize the quasi-periodic physico-chemical properties of multi-transmembrane proteins and it positively identified GPCRs at a level greater than 90% even with segments as short as 100 amino acids.
Journal of Machine Learning ResearchMultiple instance learning of real valued data
59 Citations2003Daniel R. Dooly, Qi Zhang +2 more
Extensions of k-nearest neighbors (k-NN), Citation-kNN, and the diverse density algorithm for the real-valued setting are presented and their performance on Boolean andreal-valued data is studied.
Approximating hyper-rectangles
50 Citations1997Peter Auer, Philip M. Long +1 more
Improved upper bounds for a class of such problems of " approximating " highdlmensional rectangles that arise in PAC learning and pseudorandomness are presented.
Machine LearningPAC Learning Axis-aligned Rectangles with Respect to Product Distributions from Multiple-Instance Examples
39 Citations1998Philip M. Long, Lei Tan
A polynomial-time algorithm for learning axis-aligned rectangles in Qd with respect to product distributions from multiple-instance examples in the PAC model that achieves accuracy with probability 1-δ in O.
Journal of Computer and System SciencesAgnostic Learning of Geometric Patterns
38 Citations2001Sally A. Goldman, Stephen Kwek +1 more
This paper presents an efficient online agnostic learning algorithm for learning the class of constant-dimensional geometric patterns, which are more complex than any class of geometric patterns previously studied.
Annals of Mathematics and Artificial IntelligenceMultiple-Instance Learning of Real-Valued Geometric Patterns
26 Citations2003Sally A. Goldman, Stephen Scott
This work defines and study a real-valued multiple-instance model in which each multiple- instance example (bag) is given areal-valued label in [0, 1] that indicates the degree to which the bag satisfies the target concept.
A novel bag generator for image database retrieval with multi-instance learning techniques
23 Citations2004Zhi-Hua Zhou, Min-Ling Zhang +1 more
A novel bag generator named ImaBag is presented, where the pixels of each image are first clustered based on their color and spatial features and then the clustered blocks are merged and converted into a specific number of instances.
An extended kernel for generalized multiple-instance learning
23 Citations2005Qing Tao, Stephen Scott +3 more
It is found that the new kernel generalized noticeably better than the old one in content-based image retrieval and biological sequence analysis and was slightly better or even with the old kernel in the other applications, showing that an SVM using this kernel does not overfit despite its richer representation.
Journal of Robotic SystemsVisual control of autonomous mobile robot based on self‐organizing model for pattern learning
22 Citations1988Hisashi Suzuki, Suguru Arimoto
A self-organizing model for pattern learning together with an application to an autonomous mobile robot system that can universally learn and reproduce any pattern of input-output response desired is proposed.
Machine LearningA Theoretical and Empirical Study of a Noise-Tolerant Algorithm to Learn Geometric Patterns
14 Citations1999Sally A. Goldman, Stephen Scott
An efficient noise-tolerant algorithm (designed using the statistical query model) to learn the class of one-dimensional geometric patterns and an empirical study of the algorithm provides some evidence that statistical query algorithms may be valuable for use in practice for handling noisy data.
Machine LearningPAC Learning of One-Dimensional Patterns
9 Citations1996Paul W. Goldberg, Sally A. Goldman +1 more
This work relates the concept class of geometric patterns to the landmark matching problem and presents a polynomial-time algorithm that PAC-learns the class of one-dimensional geometric patterns.
Journal of Computer and System SciencesOn approximating weighted sums with exponentially many terms
9 Citations2004Deepak Chawla, Lin Li +1 more
Applications of Markov chain Monte Carlo (MCMC) methods to estimate the total weight ofMultiplicative weight-update algorithms such as Winnow and Weighted Majority are explored and empirical results are presented indicating that in practice, the time complexity is much better than what is implied by the worst-case theoretical analysis.
Learning one-dimensional geometric patterns under one-sided random misclassification noise
7 Citations1994Paul W. Goldberg, Sally A. Goldman
This work relates the concept class of geometric patterns to the landmark recognition problem and presents a polynomial-time algorithm that PAC-learns the class of one-dimensional geometric patterns when the negative examples are corrupted by a large amount of random misclassification noise.
Lincoln (University of Nebraska)On Modeling Protein Superfamilies with Low Primary Sequence Conservation
4 Citations2003Stephen Scott, Hongbing Ji +3 more
This work presents several algorithms for identifying thioredoxin (Trx)-fold proteins containing a conserved CxxC motif (two cysteines separated by two residues) in the multipleinstance learning model.